Wiener Filter Realization for Target Detection Using Group Delay Statistics

Abstract

One of the most commonly used statistical signal processing functions is least mean square linear filtering, i.e., “Wiener filtering.” However, the Wiener filter transfer function is based on the assumption that accurate statistical models are available for relevant signal and noise phenomena, which limits its application to many practical situations. In addition, ill-designed transfer functions can cause the filtering to be unreliable. Based on the concept of group delay statistics, a new technique is presented in this paper for Wiener filter realization of the target detection problem. This technique has been verified both by simulation and ultrasonic flaw detection experiments, yielding significant signal-to-noise ratio enhancement. © 1993 IEEE